The three major AI models do not share one opinion. This page maps where they genuinely part ways about the same brand in the same category: each model's favorites, its blind spots, and the brands they split on hardest.
Temperaments. ChatGPT runs warm (102 favorites, 62 blind spots). Claude runs warm (140 favorites, 94 blind spots). Gemini runs cold (96 favorites, 120 blind spots). The closest pair is ChatGPT and Gemini (typical gap 11 points); Claude is the odd one out.
Where the fights are. AI Visibility & AEO (18), Dental Lab Management (17), Museum & Cultural Institution (16), Veterinary Practice Software (15), Medical Practice Management & EHR (13) lead the disagreement count. Fittingly, the models fight hardest over who is good at AI visibility itself.
Why they disagree. About 81% of these are awareness gaps: the cold model barely names the brand at all. The rest are preference gaps, where the brand is named just as often but ranked lower.
Brands ChatGPT scores at least 20 points above (or below) the average of the other two models in the same category. Showing the strongest 10 each way of 102 favorites and 62 blind spots.
Rates far higher than the others
Rates far lower than the others
Brands Claude scores at least 20 points above (or below) the average of the other two models in the same category. Showing the strongest 10 each way of 140 favorites and 94 blind spots.
Rates far higher than the others
Rates far lower than the others
Brands Gemini scores at least 20 points above (or below) the average of the other two models in the same category. Showing the strongest 10 each way of 96 favorites and 120 blind spots.
Rates far higher than the others
Rates far lower than the others
The brands the three models disagree on hardest: highest minus lowest per-model score for the same brand in the same category.
| Brand | Category | ChatGPT | Claude | Gemini | Spread |
|---|---|---|---|---|---|
| PayThePoolMan | Pool Service Management | 38 | 0 | 73 | 73 |
| Evident | Dental Lab Management | 0 | 0 | 71 | 71 |
| Osiris | Funeral Home & Cemetery Management | 0 | 68 | 15 | 68 |
| Pool Brain | Pool Service Management | 65 | 17 | 0 | 65 |
| CareStack | Dental Practice Management | 36 | 4 | 66 | 63 |
| FACTS | K-12 School Management (SIS) | 63 | 0 | 25 | 63 |
| QS/1 | Pharmacy Management | 15 | 63 | 0 | 63 |
| Profound | AI Visibility & AEO | 75 | 60 | 12 | 62 |
| LendingPad | Mortgage Lending & Origination | 39 | 1 | 63 | 62 |
| Clinicient | Physical Therapy Practice | 25 | 65 | 3 | 62 |
How to read this: each per-model score comes from 24 answers (8 prompts asked 3 times) in the September 2026 snapshot, scored 0-100 the same way as the main index. Honest error bars: resampling the underlying answers moves a per-model score by about 8 points, so the 20-point bar is roughly twice the sampling wobble, and the top tenth of all divergences. Entries near the bar can still be sampling luck; the giant splits in the table below cannot. The real confirmation is time: divergences that persist across monthly snapshots are bias, ones that vanish were noise, and this page re-measures every month. Blind-spot tags: "rarely names it" means the cold model's mention rate is at most 60% of its peers' (an awareness gap); "names it, ranks it low" means the brand comes up as often but places worse (a preference gap). A brand on this page is not better or worse; the models simply disagree about it, and if you are that brand, the model that is cold on you is a channel where buyers are not hearing your name. Full recipe on the methodology page.